Reference- Dependence in Dutch Football

Size: px
Start display at page:

Download "Reference- Dependence in Dutch Football"

Transcription

1 ERASMUS UNIVERSITY ROTTERDAM Erasmus School of Economics Master Thesis Reference- Dependence in Dutch Football Outcome Uncertainty and Stadium Attendance Author: Martijn Oeij Student number: Supervisor: Prof. dr. Han Bleichrodt Co-evaluator: Timo Lambregts

2 Abstract The purpose of this thesis is to look into the relationship between uncertainty of the outcome of Dutch football games and the stadium attendance of those particular games. Two main theories are proposed by previous literature as to what this relationship looks like. The classic uncertainty of outcome hypothesis (UOH) suggests that more uncertainty yields more stadium attendance. A model with reference-dependent preferences coupled with loss aversion proposes the opposite, i.e. less uncertainty yields more stadium attendance. Three Dutch football clubs provided attendance data for the analysis of this thesis, namely FC Groningen, Heracles and PSV. Betting odds are used to calculate the uncertainty of outcome. The estimated random effects tobit model provides evidence in favour of the UOH. The maximum stadium attendance is achieved at the uncertainty level where the home team is three times as likely to win as the away team. Other significant influences on stadium attendance are the away team, the league rank of the away team, derby matches, renovation of the stadium, the day of the game, the month in which the game is played, and the season. 2

3 Acknowledgement I would first like to thank my master thesis supervisor prof. dr. Han Bleichrodt of the Erasmus School of Economics. Prof. Bleichrodt helped me throughout the challenging process of writing this thesis. He steered me in the right direction whenever I had trouble with anything. Data collection was perhaps the most challenging part and prof. Bleichrodt supported me with approaching football clubs. Despite the fact that a relatively small number of football clubs were willing to provide data, prof. Bleichrodt acknowledged my enthusiasm for the subject and allowed me to pursue my own interest with great freedom in my approach. I would also like to thank the football clubs that were willing to participate in the research of this thesis. In particular thanks to Coen Schroots from PSV, Martijn van der Lee from Heracles, and Richard van Elsacker from FC Groningen. In a busy Eredivisie season they found the time and willingness to provide me with data. 3

4 Table of contents Abstract... 2 Acknowledgement Introduction The model The Uncertainty of Outcome Hypothesis Loss aversion Application to football Econometric model Literature Conclusions from the literature Data & Methods Data Methods Results Conclusion Limitations References

5 1. Introduction Uncertainty can be an important component of what makes a sporting event interesting. A team or athlete never has a 100 per cent probability of winning and this is part of the fun. In sports economics, extensive research has been done that looks into the relationship between uncertainty in the outcome of a sports game and the demand for attending this particular game. The uncertainty of outcome hypothesis (UOH) states that in a particular game uncertainty about the outcome influences the demand for the sports event in question. More specifically, the UOH argues that more uncertain games attract more consumers. The literature investigating the UOH dates back to a 1956 paper by Rottenberg (1956). He argues that an equal distribution of playing talent across Major League Baseball teams grants uncertainty of outcome, and uncertainty of outcome or competitive balance is the reason consumers pay admission fees to see these teams play. Neale (1964) also acknowledges that in sports the excitement comes from changes in the league standings of sports teams, and that more frequently changing league standings would lead to more gate receipts. He adds an insightful paradox in favour of the UOH, which is called the Louis-Schmelling Paradox, named after two former boxers. This paradox describes the situation in which one team or athlete has a monopoly in the market. A traditional firm would like to have a monopoly for its maximising profits. However, in sports this is different. If, for instance, the New York Yankees would contract all the good players from other American baseball teams, this would mean that these other teams would fail in establishing a (competitive) team. In Neale s (1964) words this would mean no games, no gate receipts, no Yankees. The Louis-Schmelling paradox is therefore a striking example to show that even when a sports team wishes to become the absolute best, they still need competition in order to generate revenue. It took some years before researchers actually tried to find empirical evidence for the UOH. An extensive literature review by Coates et al. (2014) provides a clear overview of the various articles that discuss the UOH in more debt. The articles in question differ in what sport they investigate, as well as their modelling approach. Australian football, football (soccer), American football, baseball, rugby and hockey are common sports where research is done exploring the UOH. The main conclusion from Coates et al. (2014) is that the UOH is controversial. The authors develop a model with reference-dependent preferences and with their own data from the Major League Baseball, the authors show that the UOH does not hold. In fact, games with less uncertainty yield higher attendance figures. The model with reference-dependent preferences developed by Coates et al. (2014) is relatively new and can be applied to other sports. In short, reference-dependent preferences combined with loss 5

6 aversion predict an opposite relation between uncertainty of the outcome and attendance of a sports match, compared to the UOH. Therefore, it is interesting to see how this reference-dependent preference model can help in identifying the UOH in football (soccer). Research is done on the topic of UOH and European football leagues. However, none have looked into Dutch football. This thesis will focus on the highest Dutch national football league, called the Eredivisie. The subject of competitive balance is still being looked at in the Dutch leagues. In May 2017 Dutch football clubs rejected several proposals for reforming the league (NU, 2017), including a proposal to reduce the number of participating clubs in the highest division. One of the arguments for reforming the Eredivisie is competitive balance and the fact that this would yield a higher attendance rate. However, results from previous studies regarding the UOH are mixed and therefore research needs to be done to see if competitive balance is indeed an argument for reforming the league to attract more followers. This leads to the following research question: Does uncertainty in the outcome of a match increase or decrease the stadium attendance for clubs in the Dutch Eredivisie? To get insight into this topic, the UOH model with reference-dependent preferences is outlined and explained in detail. Secondly, the relevant literature is discussed. Finally, the model is applied to data from several clubs playing in the Eredivisie. With betting odds and attendance data from several Dutch football clubs the relationship between the probability of winning and live game attendance is analysed. The following sections elaborate on these research steps. 2. The model Coates et al. (2014) developed a model for investigating in which way uncertainty in the outcome of a particular sports game influences stadium attendance. They extend the model by Card and Dahl (2011), who look at unexpected outcomes in American football and the relationship with family violence. Coates et al. (2014) further optimize this model to investigate the UOH. This model looks exclusively at individual match level uncertainty of outcome and this will also be the focus in this thesis. The model is based on the assumption that a game has two outcomes, namely a win or a loss. In the remainder of the explanation of the model this assumption is extended to football, while in reality a draw is a common outcome. In the dataset used in this thesis 20.53% of the games ended in a draw. This issue is discussed later in this section. According to standard consumer theory, consumers who attend a game receive consumption utility. The utility gained from a win is given by U w and the utility gained from a loss is U L, where 6

7 U W > U L is assumed. The objective probability of winning a game for the home team is given by p, where 0 p 1. Expected utility from attending a game for home supporters is then given by E[U] = pu W + (1 p)u L. (1) In this standard consumer theory model expected utility increases with the probability of a home win. Following this model it is expected that more successful teams have higher attendance. However, this is not consistent with the UOH. Furthermore, from prospect theory it is known that reference points play an important role in decisions under risk (Kahneman & Tversky, 1979). Therefore, the standard consumption utility model needs an extension. Koszegi and Rabin (2006) call this extension gain-loss utility, which is the part of the utility function that examines the relationship between the actual outcome of an event and the reference point for that particular event. Each event has its own reference point. Assume that the outcome of a football match y equals 1 if the match is won by the home team and 0 if the home team loses. The reference point for the home fan for the football game is given by E(y = 1) = p r. Deviation from this reference point generates gain-loss utility, according to the model developed by Koszegi and Rabin (2006). The marginal impact of this deviation is α if the home team wins, where α > 0. The utility for home fans if their team wins is given by U W + a(y p r ) = U W + α(1 p r ). (2) From this equation it can be seen that a positive deviation from the reference point results in an increase in total utility, i.e. an unexpected win results in higher utility compared to an expected win. The marginal impact in the gain-loss utility function when the home team loses is given by β, where β > 0. The utility for home fans if their team loses is given by U L + β(y p r ) = U L + β(0 p r ). (3) This equation shows that a negative deviation from the reference point results in a decrease of total utility, i.e. an expected loss generates higher utility compared to an unexpected loss. Figure 1 provides a graphical illustration on the relationship between the match outcome, reference points and the utility generated by those two factors. It can be seen that the highest utility is achieved when p = 0, i.e. the home team is perceived to have 0 per cent probability of winning, and the home team wins. Likewise, the lowest utility is generated when p = 1, i.e. the home team is perceived to have 100 per cent probability of winning, and the home team loses. These two extreme cases are the scenarios in which the deviation from the reference point in the gain-loss utility is at its maximum. 7

8 U U W + α slope = α U W U L slope = β U L β p r = 1 p r Figure 1 Utility of game outcomes with references Now assume the reference point for the home fans equals the objective probability that their team wins. Utility from attending a game is given by E[U] = p[u W + α(1 p)] + (1 p)[u L + β(0 p)]. (4) After rearranging the following equation is obtained E[U] = (β α)p 2 + [(U W U L ) (β α)]p + U L. (5) This utility equation incorporates both consumption utility (equation 1) and gain-loss utility and shows a quadratic function of the probability that the home team wins. Consumers use the utility function from equation 5 in deciding whether or not to attend a game. They compare the expected utility from equation 5 to a reservation utility v that they get when not attending a game, i.e. the reservation utility represents other leisure activities. This reservation utility v has a distribution of [v, v]. The most determined fans will have low reservation utility close to v, while casual fans have a higher reservation utility. This means that for casual fans a higher expected utility is needed before they decide to attend a game. 8

9 2.1 The Uncertainty of Outcome Hypothesis E[U] v U W U L p 0 p max p 1 1 p Figure 2 Classic UOH, concave function of probability As named by Coates et al. (2014), the classic UOH predicts a concave relationship between the probability of the home team winning and the expected utility of attending a game. This is illustrated in figure 2. The maximum p max is reached at the probability interval (0.5, 1) of the home team winning, likely around 55 per cent as hypothesized by Rottenberg (1956). In figure 2 it is easy to see that as the outcome becomes more uncertain, i.e. the probability of winning gets closer to p max, more consumers along the distribution of reservation utility [v, v] will decide to attend a game. In the case presented in figure 2, the consumer will decide to attend the game if p 0 < p < p 1. Concavity of the expected utility function is achieved if in equation 5 (β α) < 0 and [(U W U L ) (β α)] > 0. The first condition (β α) < 0 that needs to hold for the classic UOH states that the marginal impact of a positive deviation from the reference point is greater than the marginal impact of a negative deviation from the reference point. In other words, an unexpected win should result in higher marginal utility than an unexpected loss. The second condition [(U W U L ) (β α)] > 0 is logical, given that (β α) < 0 and U W > U L. 9

10 Using the first derivative of equation 5 it is possible to define p max. The first order condition is defined as follows p max = 1 2 (UW U L ) 2(β α) (6) where, given that (β α) < 0, it can be shown that p max > 1. Also the classic UOH is subject to 2 p max < 1 by definition, which combining with equation 6, shows that (U W U L ) < (α β). This relationship shows that a consumer prefers more uncertain games over their team winning for sure. This is exactly what the classic UOH hypothesizes and in this reference-dependent preference model this is the way in which the classic UOH is justified. 2.2 Loss aversion Contradictory to the UOH is a situation in which more certain games are preferred over uncertain games. This case can best be described by loss aversion. As Kahneman and Tversky put it: losses loom larger than gains (Kahneman & Tversky, 1979). In other words, losing results in a bigger decrease in utility than a gain leads to an increase in utility. In the sports framework presented in this thesis this would translate into β > α, meaning that the marginal impact of a negative deviation from the reference points is larger than the marginal impact of a positive deviation from the reference point. Instead of the concavity of the classic UOH utility function, the utility function under loss aversion is convex, as is shown in figure 3. In order to separate the consumption utility and the gain-loss utility in the case of loss aversion, equation 4 is rearranged into E[U] = [pu W + (1 p)u L ] + (α β)p(1 p). (7) The consumption utility [pu W + (1 p)u L ] increases with p. The gain-loss utility (α β)p(1 p) first decreases with p until p = 1, then increases with p, given that (α β) < 0 for loss aversion to 2 hold. The lowest utility is achieved at p min and the first order condition is now p min = 1 2 (UW U L ) 2(β α), which is similar to equation 6. If p < p min then the negative effect of the gain-loss utility outweighs the positive effect by the consumption utility. When p > p min the opposite is the case. In short, with reference-dependent preferences and loss aversion, attending a game that is either an expected win or an expected loss, generates more utility than games with more uncertainty. So, in this model an expected loss, e.g. p < 0.1, is preferred to a more uncertain match, e.g. p = 0.5. This can be motivated by the fact that casual fans are interested in seeing surprising results, i.e. David versus Goliath scenarios (Buraimo, 2014). This happens when the home team is highly expected to lose, e.g. p < 0.1, and the home team wins regardless of that probability. These surprising results are not 10

11 E[U] U W U L v p 0 p min p 1 1 p Figure 3 Reference-dependent preferences and loss aversion explained in the classic UOH framework, while in reality it could be the case that some fans seek games to attend at which they can witness a potential surprising result. Also, this version of the model allows for explaining the fact that fans would only want to see their team win, something the classic UOH fails to recognise. In figure 3 p 0 and p 1 are the cut-off points for which attending the live game is more attractive than not, given reservation utility v. In the case of loss aversion the reservation utility is important. If for some consumers the reservation utility is relatively high, then v > U L. In this situation p 1 is at a relatively high level and therefore identification of the utility function with loss aversion is empirically difficult. The reason for this is that in the situation where v > U L, observations where p < p 1 may show a flat relationship between utility and p. If the reservation utility is relatively low, which is assumed to be the case for strong fans of a team, then this will not be a problem. Furthermore, the difference between U W and U L is crucial. If this difference is relatively large and with U W > U L, then the expected utility function will have the same relationship to p as in equation 1, i.e. increasing in p. 11

12 2.3 Application to football This framework assumes that the outcome of a game is binary, i.e. y = 1 for a win and y = 0 for a loss. However, in football the outcome of a draw is not uncommon. In league football a win results in 3 points for the winning team and a draw results in 1 point. This complicates the model. The solution for this is as follows. Suppose fans of a strong team expect their team to win against a weak opponent. In this case even a draw for the fans of the strong team feels like a loss; they lose 2 points when their reference is winning. So for strong teams the solution in the model is to transform the binary variable into two outcomes: a win y = 1 and not winning y = 0 (draw or loss). Hereby, it is the assumption that in this case the utility of a draw and loss is the same, which is a questionable assumption. For a weak team the reasoning is the same. Fans of a weak team would like a draw against stronger opponents. Therefore, in the outcome of a draw they win 1 point. For weak teams the binary variable y becomes 1 when winning or drawing, and 0 when losing. These assumptions are open for debate, because it is questionable to assume for fans of a strong team the utility generated by a draw equals the utility generated by a loss. However, for empirically looking at data and interpreting the results, this does not matter. For the reference-dependence models with either classic UOH or loss aversion to hold, the relationship between the probability of winning and stadium attendance remains the same, despite the fact that a third outcome could be added to the model. Also, for the analysis of data in the following sections, the fact that the draw outcome complicates the theoretical model does not affect the implications or interpretations of the results. In the literature section different studies investigating football are mentioned which all incorporate a form of p. The authors from these studies all compare their results to the same classic UOH as authors from other sports. 2.4 Econometric model Looking at the expected utility equation 5 it is possible to simplify the terms as a function of the probability of the home team winning. This looks like E[U] = γp 2 + θp + λ. (8) Now it is possible to derive an econometric model (Coates et al., 2014) explaining attendance at live games. Therefore, it is assumed that in the observations the attendance of a game depends on the number of people that have higher expected utility of attending a game compared to their reservation utility, so E[U] > v. The econometric model with home team i and away team j at time t is as follows ln Attendance ijt = λ + γp 2 ijt + θp ijt + f(covariates) + ε ijt (9) 12

13 where f(covariates) are parameters for characteristics of the home team, the away team, the match itself, and time variables, and ε ijt is the random error. The parameters of interest are γ, θ and λ, where γ = β α, θ = (UW U L ) (β α) v v v v and λ = UL v according to the framework developed in the v v previous sections. With this econometric model it is now possible to look at other literature for results concerning the influence of the probability of winning for the home team on live attendance. In order to do this in a well-ordered way, it is useful to distinguish between the different scenarios and match these scenarios with the reference-dependent preference models with classic UOH or loss aversion. The following hypotheses are therefore needed (Coates et al., 2014): H1 Loss aversion is supported by the model: γ > 0, i.e. β > α. H1a γ > 0 and θ < 0, i.e. β > α and (U W U L ) < (β α). This means that for the marginal consumer the marginal impact of loss aversion is larger than the consumption utility difference between a home win and a home loss, given that U W > U L. H1b γ > 0 and θ > 0, i.e. β > α and (U W U L ) > (β α). This means that for the marginal consumer the marginal impact of loss aversion is smaller than the consumption utility difference between a home win and a home loss, given that U W > U L. H2 The marginal consumer does not behave according to reference-dependent preferences and the utility for a home win is larger than the utility for a home loss: γ = 0 and θ > 0, i.e. β = α and U W > U L. H3 The classic UOH is supported by the model: γ < 0 and θ > 0, i.e. (β α) < 0 (U W U L ). This means that the marginal consumer has a preference for uncertain games and gets more utility from an unexpected win compared to an unexpected loss. With these structured hypotheses it is possible to look at the existing literature, and conduct an analysis with data from Dutch football clubs. 3. Literature An overview of the relevant literature that investigates the UOH in football leagues is presented in table 1. These papers all include a measure for outcome uncertainty as an independent variable and stadium attendance as the dependent variable. Furthermore, most papers use a similar approach as to what covariates to use in the estimation of the econometric model. These covariates often include variables for quality and form of both teams, and certain incentive variables, such as the stage of the competition, whether a game is considered a derby, distance between the two teams, and time, day 13

14 and month of the game. In the remainder of this section these papers are discussed more elaborately. The first notable papers are by Peel and Thomas (1988; 1992). They investigate the UOH in the top four divisions of the English Football League 1. In their 1988 paper they include only a linear measure for outcome uncertainty, namely the probability of winning for the home team, derived from betting odds. Also, they employ data on the position of the home/away team, the distance between the home and away team, a variable for derby matches, and a variable indicating what stage the competition is at. With these parameters they estimate an ordinary least squares (OLS) model, which is a relatively simple way of looking at the UOH, compared to the literature that is later discussed. Peel and Thomas (1988) find with this simple model that their single variable for home team probability of winning is positive and significant for all four divisions separately, meaning that Author(s) Football league Uncertainty measure Results Support Peel and Thomas (1988) English tier Betting odds θ > 0 H2 Peel and Thomas (1992) English tier Betting odds θ < 0, γ > 0 H1a Czarnitzki and Stadtmann (2002) German top tier Betting odds θ = 0, γ = 0 - Forrest and Simmons (2002) English tier Betting odds - H3 Forrest et al. (2005) English tier Betting odds - H3 Falter, Pérignon, & Vercruysse, French top tier f(points) θ = Buraimo & Simmons (2008) English top tier division Betting odds θ < 0, γ > 0 H1a Benz et al. (2009) German top tier Betting odds, θ > 0, f(win%) γ < 0 H3 Madalozzo & Berber Villar, 2009 Brazilian top tier f(win%) θ = 0 - Buraimo and Simmons (2009) Spanish top tier division Betting odds θ < 0, 2007 γ > 0 H1a Buraimo (2014) English tier Betting odds θ < 0, γ > 0 H1a Reilly (2015) Irish top tier Betting odds - - Martins and Cró (2016) Portuguese top tier Betting odds θ < 0, γ > 0 H1a Jena and Reilly (2016) Irish 2 nd tier Betting odds θ > 0, γ < 0 H3 Table 1 Literature of outcome uncertainty and live game attendance in football 1 The well-known English Premier League was formed in 1992, so before 1992 the top tier of the English Football League was the highest division in English football. 14

15 attendance is higher when the home team has a greater chance of winning (H2). Their subsequent paper further develops this simple model, with two extensions, namely a variable for the square of the probability of the home team winning, and a measure of the attendance of the preceding match for the home and away team. Using, once again, an OLS model, the results by Peel and Thomas (1992) suggest a relationship between live game attendance and outcome uncertainty that supports H1a. Czarnitzki and Stadtmann (2002) investigate the relationship between match uncertainty, calculated using betting odds, and live game attendance in the Bundesliga, which is the highest division in German football. Additionally, they include a variable for uncertainty in the outcome of the whole season through a measure that compares the points needed to win the Bundesliga that season to the points a certain team already has. Other variables include the reputation of a team, the number of supporter clubs, the size of the market, the form of the teams, weather conditions, the stage of the competition, and a dummy indicating if the match was broadcasted live on television. The measure for the reputation of a team is computed using the final rankings of a certain team over a period of 20 years. The authors argue that the attendance data is right censored, i.e. due to the capacity constraint of each stadium, 25% of the games were sold out and therefore true demand exceeded the maximum capacity of the stadium. This is the reason that Czarnitzki and Stadtmann (2002), and other authors mentioned later in this section, use a tobit model. Their results show coefficients that would support H1a. However, these coefficients were not statistically different from zero. The same applies to their uncertainty of outcome measures on the seasonal level. Therefore, Czarnitzki and Stadtmann (2002) conclude that the reputation of a team and the stage of competition are more important factors in explaining attendance. Forrest and Simmons (2002) employ a similar approach as Peel and Thomas (1988; 1992) with regards to the model specification. However, they have concerns about the ability of the betting market to set betting odds unbiasedly. The authors therefore look into possible biases in betting odds. One bias that is backed by previous research is the favourite-longshot bias (Cain et al., 2000). The favourite-longshot bias states that favourites win more often than the subjective market probabilities (set by the bookmakers) imply, and longshots less often (Cain et al., 2000). Furthermore, Forrest and Simmons (2002) correct the betting odds for differences in mean attendances of the two teams playing, since the authors suspect that bookmakers vary their betting odds according to the relative level of support for the two teams. After correcting these biases they find evidence in favour of H3, using data from the English Football league excluding the English Premier League. Contrary to estimating a model for each division, fixed effects for each team are used to capture heterogeneity between the teams. Instead of the conventional functional form of 15

16 the winning probability for the home team, Forrest and Simmons (2002) use the ratio between the probability of winning of the home team and the probability of winning for the away team as the uncertainty measure. With this uncertainty measure the classic UOH is supported in 97.8% of the football games the authors analysed. Forrest et al. (2005) also find evidence in favour of H3 with similar data and methodology. The study by Falter et al. (2008) has its focus primarily on the effect of a World Cup victory of the national football team on the attendance demand in that country s domestic football league in the period after the World Cup victory. This research is somewhat different in terms of the main research question, but in their attendance demand model a measure for match uncertainty is included, which makes this study worth mentioning. Instead of using betting odds, Falter et al. (2008) use a selfcomputed measure of outcome uncertainty. It takes into account the points of the home team, points of the away team, and the home advantage. Their model explaining attendance includes variables of the World Cup victory effect, outcome uncertainty, stage of competition, quality of both teams and the match, team fixed effects, weather variables and a variable indicating if the match was broadcasted on live television. Their measure for outcome uncertainty proves to be insignificant in the OLS estimation for the French dataset that is used. Madalozzo and Berber Villar (2009) also investigate the effect of outcome uncertainty on match attendance without the use of betting odds. The authors estimate fixed and random effects panel data models using data from the highest division in Brazil including variables on the quality of both teams and the match, cost of attending the match, stage of the competition and outcome uncertainty. Madalozzo and Berber Villar (2009) define outcome uncertainty in four variables; the difference in rankings between the home team and the away team, a probability measure for being the league leader, a probability measure of going to the Libertadores Cup 2, and a probability measure of leaving the rankings associated with relegation. For the uncertainty measures, the probability of being the league leader and probability of leaving the relegation zone were significant. Also the stage of competition had a significant influence on attendance. The approach by Madalozzo and Berber Villar (2009) use for their uncertainty measures is different compared to the model that is developed in this thesis. However, it is interesting to note that for instance the probability of being the league leader significantly increases game attendance. The first cited study that looks into attendance demand in arguably the most popular league in the world (Curley & Roeder, 2016), the English Premier League, is a paper by Buraimo and Simmons (2008). Using betting odds as the basis for the uncertainty measure they estimate a tobit model, with 2 The Latin American Football Cup, comparable to the UEFA Champions League. 16

17 comparable variables as the studies cited above. The results from this study are in line with H1a. A year later Buraimo and Simmons (2009) focused on another acclaimed football league, namely the Spanish Primera Division. For this study they estimate with similar variables a Prais-Winsten panel regression model, in which error terms are correlated across panels, i.e. home teams. Their results for stadium attendance are similar to their 2008 paper and support H1a. Interestingly, Buraimo and Simmons (2009) also look at the relationship between outcome uncertainty and the television viewer ratings for a particular match. They find evidence in support of the classic UOH with regards to television viewers. In response to studies using OLS or tobit estimations, Benz et al. (2009) investigate demand for game attendance in the German Bundesliga with a distinctive approach. They employ a quantile regression model to overcome the assumption that average effects of the regressors apply to the whole distribution of demand. An advantage of quantile regression is that the importance of influence factors on attendance varies with the level of this attendance, which might be useful if for instance the home supporter s utility of their team winning is increasing in the number of home supporters. In this case, with quantile regression the probability of the home team winning has a greater influence for larger quantiles of attendance. Their data includes variables on the quality of both teams, quality of the match, economic factors and weather variables. Furthermore, they use a variety of measures for outcome uncertainty. These outcome uncertainty measures include the difference in rankings between the home team and the away team, a measure that compares the points per game for both teams and corrects for home advantage, and probabilities based on betting odds. Their results suggest evidence in support of H3. However, the coefficients for the uncertainty of outcome measures were only significant on a 10% significance level. Additionally, Benz et al. (2009) conclude that uncertainty of outcome is only a second-order influence factor and that the reputation of a team is statistically more important in explaining attendance demand. The latter is also what Czarnitzki and Stadtmann (2002) found in their study of the Bundesliga. Buraimo (2014) elaborated on his earlier research (Buraimo & Simmons, 2008; Buraimo & Simmons, 2009) by examining the five highest football divisions in England 3. Similarly to Buraimo and Simmons (2008), the author estimates a tobit model for the English Premier League, since 47% of the games in this league are sold out in the dataset he uses. For other English leagues the author uses a panel data model with fixed effects. With the commonly used variable types each of the five leagues is analysed separately. Buraimo (2014) finds in each league results that confirm the reasoning of H1a. 3 This includes the English Premier League, the English Football League Championship, the English Football League One, the English Football League Two, and the National League. 17

18 Reilly (2015) investigates the relationship between outcome uncertainty and live game attendance in the highest division in Ireland. Irish football is relatively small in terms of attendance compared to the leagues discussed above; in 2014 the average attendance was (Reilly, 2015). The author estimates, with conventional variables, different statistical models. The first notable model is comparable to the models that the other cited articles employ. In this fixed effects panel data model a linear and quadratic parameter is used for the probability of winning of the home team. This model confirms H1a. However, Reilly (2015) argues that the use of splines provides a more clear representation of the data. He uses two splines for the probability of winning of the home team in order to recreate the U-shaped relationship between probability of winning and match attendance as hypothesized by H1a. Of these two splines, only the first one is statistically significant. This implies that a negative relation exists between the probability of winning of the home team in the range of and , and match attendance. Outside of this range of the win probability no significant influence on attendance could be found. In other words, matches where the home team has a probability of winning smaller than 0.25, there is less attendance the larger the probability of winning. The author concludes that the latter model dominates the quadratic model in terms of goodness-of-fit. Following the paper by Reilly (2015) is the paper by Jena and Reilly (2016), which examines the even smaller 4 second division in Ireland. The authors now choose to focus on a quadratic function of the probability of winning for the home team and use the conventional variables. Here they find evidence in support of the classic UOH (H3). The last cited paper is a study by Martins and Cró (2016), which explores the highest Portuguese division. Next to the conventional variables, the authors use a variable for uncertainty of outcome on the seasonal level. Also, they argue that their variable for television broadcasted matches is subject to endogeneity, because the most attractive matches are chosen to be broadcasted. Therefore, Martins and Cró (2016) estimate a two-stage tobit model, using a time indicator as an instrumental variable for the television broadcast variable. The results obtained from this model support H1a. 3.1 Conclusions from the literature Clearly, there is no consensus on the proposed relationship between match uncertainty and attendance in football. Coates et al. (2014) argue that this could be due to the fact that the specifications of the variables differ in these various papers. Furthermore, as is seen from the model developed in this thesis, the coefficients for the parameters of the probability of the home team winning reflect both the consumption utility (U W and U L ) and the marginal utility of a win or a loss compared to the reference point (α and β). Cultural differences in the consumption utility might be 4 Mean attendance in was (Jena & Reilly, 2016). 18

19 an important factor in the cited research. Also, consumption utility of sports might not be constant over time, since interest in sports varies over time, e.g. Falter et al. (2008) show that demand for football in a country rises if the national team of that country performs well. Consequently, it might be the case that studies from other countries and time periods are not externally valid and therefore it is interesting to conduct a study looking specifically at the Dutch Eredivisie with recent data. 4. Data & Methods 4.1 Data The Dutch Eredivisie is the highest division in professional football in the Netherlands and comprises 18 teams. Within one season all teams play against each other twice; one home game and one away game. Points are awarded based on the outcome of the game. In case of one team winning, the winning team receives 3 points and the losing team receives no points. When two teams draw, both receive 1 point. The league rankings are based on the total amount of points each team has gathered throughout the season and at the end of the season the ranking of a team has certain consequences. Rewards for the highest ranking teams include winning the Eredivisie as champion or qualifying for (playoffs for) next season s European competitions. Negative consequences face the lower ranking teams, namely the risk of relegation to a lower division. Attendance Data Attendance data was collected from each club individually. One criterium for clubs to be considered was if their team played in the Eredivisie for most of the recent years. Clubs were contacted and asked if they were willing to participate. FC Groningen, Heracles and PSV were willing to participate with this research. Table 2 provides an overview of the attendance data that is used in this thesis. As can be seen from table 2, the number of observations in this dataset differs between clubs. The reason for this is that some clubs were not able to generate more data. The data gathered by clubs include total tickets sold for each home game, total season tickets sold that season and the maximum capacity of the stadium. Note that this data includes the total tickets sold and not total attendance on the day of the match. Data on the latter is not readily available. Probabilities and betting odds The probability of the home team winning is calculated using betting odds. The use of betting odds provides a market based approach to evaluate the probabilities of different outcomes. Betting odds data is collected from OddsPortal.com, where betting odds from the most used online bookmakers are collected. OddsPortal.com reports the mean betting odds collected from these bookmakers. The 19

20 Club Mean Minimum Maximum Observations FC Groningen Total tickets Table 2 Descriptive statistics of attendance data. Games are sold out if the total tickets sold exceed the adjusted maximum capacity of the stadium. More information on the adjusted maximum capacity can be found in the methods section. average amount of bookmakers for each game is with a standard deviation of Online bookmakers often adjust their betting odds in the days, hours or even minutes before a match. Betting odds reported on OddsPortal.com are the most recent betting odds. Usually this is not more than a few minutes before a game. Betting odds incorporate important information on the competing teams, such as home advantage, missing players and current form. The assumption here is that betting markets are efficient and bookmakers have the incentive to be efficient in order to make profit. However, as mentioned by Forrest and Simmons (2002), some biases exist in betting markets. More recent research also confirms that backing strong favourites yields positive returns (Direr, 2013). This means that for games with a strong favourite the estimated probability of the favourite winning as constructed by the bookmaker is smaller than the actual objective probability. Due to data restrictions in this thesis it is not possible to correct the betting odds for possible biases using an equivalent approach as Forrest and Simmons (2002). Therefore in this thesis it is assumed that betting odds are indeed efficient. The probability of home team i winning the game against away team j is calculated using the formula Season tickets Single tickets Stadium capacity Sold out Heracles Total tickets Season tickets Single tickets Stadium capacity Sold out PSV Total tickets Season tickets Single tickets Stadium capacity Sold out Total Total tickets Season tickets Single tickets Stadium capacity Sold out

21 p ij = x iw 1 x iw 1 + x id 1 + x il 1 (10) where x iw, x id and x il denote the betting odds 5 for a win, draw, or loss of the home team respectively. Control variables Data on the characteristics of both teams and the match are gathered from Voetbal.com. This data includes information about the standings of both teams prior to the match, the standings of both teams of last season, pre-match points of both teams, and pre-match goals scored and conceded for both teams. Pre-match standings for the first game of the season are none-existent, which is the reason for excluding the first game of the season from the dataset. 4.2 Methods One of the aspects of football games for Dutch football clubs is the fact that a significant number of games sell out, as can be seen from table 2. This complicates the estimation of a demand function of attending a game, because real demand exceeds the maximum capacity of a particular stadium. Therefore, sold out games are censored observations. One way of taking this into account is to estimate a tobit model. Tobit regressions take into account the fact that the dependent variable might be censored, which is the case in this dataset. Tobit models are also more clear in terms of the interpretation of the estimation, compared to methods such as the quantile regression used by Benz et al. (2009). The important thing to take into consideration here is the question of what exactly constitutes a sold out game. In the Dutch league away fans are commonly seated together in a small area of the stadium. These away fans have to be segregated from the home fans because of security reasons. In some cases this leads to a number of empty seats that provide a buffer between rivalling fans, which causes the maximum capacity of the stadium to be lower. For the analysis, 5% of the maximum capacity is treated as empty seats that are due to security reasons, following the approach by Buraimo and Simmons (2008). The remaining net maximum capacity is used as the upper limit of the tobit model. This also solves the problem that in a few cases the reported total tickets sold exceeds the maximum stadium capacity by treating these few cases the same as other observations where the tickets sold exceeds the net maximum capacity. For sensitivity purposes a separate analysis is done to look at the differences in results when a different maximum capacity is used. Furthermore, the data that is used in this thesis is panel data; the home teams are the panels. 5 Betting odds are given in decimal format (EU format). 21

22 Therefore, it is most likely the case that the error terms are serially correlated because of unobserved time-constant home team characteristics. To account for this a panel data model can be estimated. This is a more efficient way of dealing with this sort of data as it exploits the correlation in the error terms. To do this the tobit model is extended to a random effects tobit model. A likelihood-ratio test confirmed that the random effects tobit model is superior to a pooled tobit model with the current dataset. The additional assumption that comes with a random effects tobit model as a panel data model is that the unobserved home team characteristics are not correlated with the independent variables. Madalozzo and Berber Villar (2009) show with their dataset on Brazilian football and a Hausman test that their random effects model is consistent and efficient, implying that unobserved characteristics are not correlated with the independent variables. This provides some evidence in favour of this assumption. From the literature review certain variables were identified as possibly relevant in the analysis of attendance demand. Equation 9 serves as the basis for the econometric model that is estimated in this thesis. The model includes most relevant variables given that it provides a sufficient fit to the data according to the AIC and BIC, and the significance of the variables. More specifically, the model for home team i, away team j at time t with random error ε ijt looks like: ln STS ijt = γp 2 ijt + θp ijt + δ 1 RankHome ijt + δ 2 RankAway ijt + δ 3 Derby ijt + δ 4 Renovate ijt + δ 5 Day ijt + δ 6 Month ijt + δ 7 Season ijt (11) + δ 8 Away ijt + ε ijt where: ln STS ijt = the natural logarithm of single tickets sold. A large number of consumers who attend a football match in the Netherlands are season ticket holders. In the dataset these season ticket holders are included in the total tickets. On average 74.51% of the total tickets sold are season ticket holders. This means that a significant part of the total tickets sold are fixed for the whole season, resulting in a rather large number of consumers to be unresponsive to any of the variables of interest. To counter this, in the analysis that follows the total tickets sold will be adjusted to reflect only the single tickets sold. This is done by subtracting the number of season tickets sold from the total tickets sold. The dependent variable for the random effects tobit model is therefore the logarithm of the single tickets sold. The transformation of the single tickets sold into a logarithm is done because the distribution of the single tickets sold is right skewed and taking the natural logarithm is a solution for this. 22

23 p ijt = the probability of the home team winning, derived from betting odds with equation 10. RankHome ijt = the rank of the home team in the Eredivisie league table prior to the match. This variable captures the current form of the home team. Fans interest might increase if their team is performing well in the current season. Also, the ranking of a team is a proxy for the relative quality of a team and fans might also be interested in seeing a quality team play. RankAway ijt = the rank of the away team in the Eredivisie league table prior to the match. This variable captures the current form of the away team. Home fans interest might also be influenced by the form and quality of the visiting team. Derby ijt = a dummy variable indicating whether the match is considered a derby match. For the analysis in this thesis a derby is arbitrarily defined as a match where two teams of close geographical proximity play each other and winning the match has significant importance for fans of both clubs regardless of the quality/form of both teams. The latter is an added definition for this thesis. The reason for this is that including certain matches in the Derby ijt variable resulted in different levels of statistical significance. For instance, PSV plays other teams that are in the same region, such as NAC Breda or Willem II. However, adding these two matches to be considered for the Derby ijt variable resulted in an insignificant parameter. Therefore, these two matches for the final analysis are excluded, with the reasoning being that for PSV fans these matches are not more important compared to other matches they play against teams of the same quality. Two matches are ultimately considered as derby matches, namely FC Groningen against SC Heerenveen, the derby of the north, and Heracles against FC Twente, the Twentse 6 derby. Renovate ijt = a dummy variable that indicates in a given season if the stadium had a renovation. This variable is included to capture any effect that the excitement about a renovation of the stadium has on the home fans for the season following the renovation. In the dataset this constitutes to one particular season for Heracles, namely the season. Day ijt = a vector of dummy variables indicating the day of the week. Controlling for the day of the week is important, because of different leisure or work-related activities are 6 Twente is a region in the east of the Netherlands. 23

Reference-dependent preferences in the case of Allsvenskan

Reference-dependent preferences in the case of Allsvenskan Reference-dependent preferences in the case of Allsvenskan Outcome uncertainty and live-game attendance Gustav Johansson Gustav Hallén Fall semester 2017 January 17th, 2018 Bachelor Thesis, 15 ECTS Department

More information

What Causes the Favorite-Longshot Bias? Further Evidence from Tennis

What Causes the Favorite-Longshot Bias? Further Evidence from Tennis MPRA Munich Personal RePEc Archive What Causes the Favorite-Longshot Bias? Further Evidence from Tennis Jiri Lahvicka 30. June 2013 Online at http://mpra.ub.uni-muenchen.de/47905/ MPRA Paper No. 47905,

More information

Is Home-Field Advantage Driven by the Fans? Evidence from Across the Ocean. Anne Anders 1 John E. Walker Department of Economics Clemson University

Is Home-Field Advantage Driven by the Fans? Evidence from Across the Ocean. Anne Anders 1 John E. Walker Department of Economics Clemson University Is Home-Field Advantage Driven by the Fans? Evidence from Across the Ocean Anne Anders 1 John E. Walker Department of Economics Clemson University Kurt W. Rotthoff Stillman School of Business Seton Hall

More information

League Quality and Attendance:

League Quality and Attendance: League Quality and Attendance: Evidence from a European Perspective* Dr Raymond Bachan* and Prof Barry Reilly ** *University of Brighton, UK **University of Sussex, UK Context Are less fashionable soccer

More information

Men in Black: The impact of new contracts on football referees performances

Men in Black: The impact of new contracts on football referees performances University of Liverpool From the SelectedWorks of Dr Babatunde Buraimo October 20, 2010 Men in Black: The impact of new contracts on football referees performances Babatunde Buraimo, University of Central

More information

Should bonus points be included in the Six Nations Championship?

Should bonus points be included in the Six Nations Championship? Should bonus points be included in the Six Nations Championship? Niven Winchester Joint Program on the Science and Policy of Global Change Massachusetts Institute of Technology 77 Massachusetts Avenue,

More information

Opleiding Informatica

Opleiding Informatica Opleiding Informatica Determining Good Tactics for a Football Game using Raw Positional Data Davey Verhoef Supervisors: Arno Knobbe Rens Meerhoff BACHELOR THESIS Leiden Institute of Advanced Computer Science

More information

University of Nevada, Reno. The Effects of Changes in Major League Baseball Playoff Format: End of Season Attendance

University of Nevada, Reno. The Effects of Changes in Major League Baseball Playoff Format: End of Season Attendance University of Nevada, Reno The Effects of Changes in Major League Baseball Playoff Format: End of Season Attendance A thesis submitted in partial fulfillment of the requirements for the degree of Master

More information

Artificial Pitches and Unfair Home Advantage in Professional Football

Artificial Pitches and Unfair Home Advantage in Professional Football TI 2017-093/V Tinbergen Institute Discussion Paper Artificial Pitches and Unfair Home Advantage in Professional Football Jan C. van Ours 1 1 Erasmus School of Economics, Erasmus University Rotterdam; Tinbergen

More information

Quantitative Methods for Economics Tutorial 6. Katherine Eyal

Quantitative Methods for Economics Tutorial 6. Katherine Eyal Quantitative Methods for Economics Tutorial 6 Katherine Eyal TUTORIAL 6 13 September 2010 ECO3021S Part A: Problems 1. (a) In 1857, the German statistician Ernst Engel formulated his famous law: Households

More information

Staking plans in sports betting under unknown true probabilities of the event

Staking plans in sports betting under unknown true probabilities of the event Staking plans in sports betting under unknown true probabilities of the event Andrés Barge-Gil 1 1 Department of Economic Analysis, Universidad Complutense de Madrid, Spain June 15, 2018 Abstract Kelly

More information

Using Actual Betting Percentages to Analyze Sportsbook Behavior: The Canadian and Arena Football Leagues

Using Actual Betting Percentages to Analyze Sportsbook Behavior: The Canadian and Arena Football Leagues Syracuse University SURFACE College Research Center David B. Falk College of Sport and Human Dynamics October 2010 Using Actual Betting s to Analyze Sportsbook Behavior: The Canadian and Arena Football

More information

Information effects in major league baseball betting markets

Information effects in major league baseball betting markets Applied Economics, 2012, 44, 707 716 Information effects in major league baseball betting markets Matt E. Ryan a, *, Marshall Gramm b and Nicholas McKinney b a Department of Economics and Quantitative

More information

Department of Economics Working Paper

Department of Economics Working Paper Department of Economics Working Paper Number 15-13 December 2015 Are Findings of Salary Discrimination Against Foreign-Born Players in the NBA Robust?* James Richard Hill Central Michigan University Peter

More information

Gamblers Favor Skewness, Not Risk: Further Evidence from United States Lottery Games

Gamblers Favor Skewness, Not Risk: Further Evidence from United States Lottery Games Gamblers Favor Skewness, Not Risk: Further Evidence from United States Lottery Games Thomas A. Garrett Russell S. Sobel Department of Economics West Virginia University Morgantown, West Virginia 26506

More information

The final set in a tennis match: four years at Wimbledon 1

The final set in a tennis match: four years at Wimbledon 1 Published as: Journal of Applied Statistics, Vol. 26, No. 4, 1999, 461-468. The final set in a tennis match: four years at Wimbledon 1 Jan R. Magnus, CentER, Tilburg University, the Netherlands and Franc

More information

Deciding When to Quit: Reference-Dependence over Slot Machine Outcomes

Deciding When to Quit: Reference-Dependence over Slot Machine Outcomes Deciding When to Quit: Reference-Dependence over Slot Machine Outcomes By JAIMIE W. LIEN * AND JIE ZHENG * Lien: Department of Economics, School of Economics and Management, Tsinghua University, Beijing,

More information

a) List and define all assumptions for multiple OLS regression. These are all listed in section 6.5

a) List and define all assumptions for multiple OLS regression. These are all listed in section 6.5 Prof. C. M. Dalton ECN 209A Spring 2015 Practice Problems (After HW1, HW2, before HW3) CORRECTED VERSION Question 1. Draw and describe a relationship with heteroskedastic errors. Support your claim with

More information

Playoff Uncertainty, Match Uncertainty and Attendance at Australian National Rugby League Matches*

Playoff Uncertainty, Match Uncertainty and Attendance at Australian National Rugby League Matches* First version: August 2010 Revised version: May 2011 Playoff Uncertainty, Match Uncertainty and Attendance at Australian National Rugby League Matches* NICHOLAS KING, P. DORIAN OWEN and RICK AUDAS Department

More information

International Journal of Computer Science in Sport. Ordinal versus nominal regression models and the problem of correctly predicting draws in soccer

International Journal of Computer Science in Sport. Ordinal versus nominal regression models and the problem of correctly predicting draws in soccer International Journal of Computer Science in Sport Volume 16, Issue 1, 2017 Journal homepage: http://iacss.org/index.php?id=30 DOI: 10.1515/ijcss-2017-0004 Ordinal versus nominal regression models and

More information

The impact of human capital accounting on the efficiency of English professional football clubs

The impact of human capital accounting on the efficiency of English professional football clubs MPRA Munich Personal RePEc Archive The impact of human capital accounting on the efficiency of English professional football clubs Anna Goshunova Institute of Finance and Economics, KFU 17 February 2013

More information

College Teaching Methods & Styles Journal First Quarter 2007 Volume 3, Number 1

College Teaching Methods & Styles Journal First Quarter 2007 Volume 3, Number 1 The Economics Of The Duration Of The Baseball World Series Alexander E. Cassuto, (E-mail: aleaxander.cassuto@csueastbay.edu), California State University, Hayward Franklin Lowenthal, (E-mail: frabklin.lowenthal@csueastbay.edu),

More information

1. Answer this student s question: Is a random sample of 5% of the students at my school large enough, or should I use 10%?

1. Answer this student s question: Is a random sample of 5% of the students at my school large enough, or should I use 10%? Econ 57 Gary Smith Fall 2011 Final Examination (150 minutes) No calculators allowed. Just set up your answers, for example, P = 49/52. BE SURE TO EXPLAIN YOUR REASONING. If you want extra time, you can

More information

What does it take to produce an Olympic champion? A nation naturally

What does it take to produce an Olympic champion? A nation naturally Survival of the Fittest? An Econometric Analysis in to the Effects of Military Spending on Olympic Success from 1996-01. Mark Frahill The Olympics are the world s greatest sporting celebrations, seen as

More information

Peer Effect in Sports: A Case Study in Speed. Skating and Swimming

Peer Effect in Sports: A Case Study in Speed. Skating and Swimming Peer Effect in Sports: A Case Study in Speed Skating and Swimming Tue Nguyen Advisor: Takao Kato 1 Abstract: Peer effect is studied for its implication in the workplace, schools and sports. Sports provide

More information

Microeconometric essays in health, sports and education economics Bojke, Christopher

Microeconometric essays in health, sports and education economics Bojke, Christopher University of Groningen Microeconometric essays in health, sports and education economics Bojke, Christopher IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you

More information

Factors Affecting Minor League Baseball Attendance. League of AA minor league baseball. Initially launched as the Akron Aeros in 1997, the team

Factors Affecting Minor League Baseball Attendance. League of AA minor league baseball. Initially launched as the Akron Aeros in 1997, the team Kelbach 1 Jeffrey Kelbach Econometric Project 6 May 2016 Factors Affecting Minor League Baseball Attendance 1 Introduction The Akron RubberDucks are an affiliate of the Cleveland Indians, playing in the

More information

Efficiency Wages in Major League Baseball Starting. Pitchers Greg Madonia

Efficiency Wages in Major League Baseball Starting. Pitchers Greg Madonia Efficiency Wages in Major League Baseball Starting Pitchers 1998-2001 Greg Madonia Statement of Problem Free agency has existed in Major League Baseball (MLB) since 1974. This is a mechanism that allows

More information

Contingent Valuation Methods

Contingent Valuation Methods ECNS 432 Ch 15 Contingent Valuation Methods General approach to all CV methods 1 st : Identify sample of respondents from the population w/ standing 2 nd : Respondents are asked questions about their valuations

More information

The Simple Linear Regression Model ECONOMETRICS (ECON 360) BEN VAN KAMMEN, PHD

The Simple Linear Regression Model ECONOMETRICS (ECON 360) BEN VAN KAMMEN, PHD The Simple Linear Regression Model ECONOMETRICS (ECON 360) BEN VAN KAMMEN, PHD Outline Definition. Deriving the Estimates. Properties of the Estimates. Units of Measurement and Functional Form. Expected

More information

Fit to Be Tied: The Incentive Effects of Overtime Rules in Professional Hockey

Fit to Be Tied: The Incentive Effects of Overtime Rules in Professional Hockey Fit to Be Tied: The Incentive Effects of Overtime Rules in Professional Hockey Jason Abrevaya Department of Economics, Purdue University 43 West State St., West Lafayette, IN 4797-256 This version: May

More information

Revisiting the Hot Hand Theory with Free Throw Data in a Multivariate Framework

Revisiting the Hot Hand Theory with Free Throw Data in a Multivariate Framework Calhoun: The NPS Institutional Archive DSpace Repository Faculty and Researchers Faculty and Researchers Collection 2010 Revisiting the Hot Hand Theory with Free Throw Data in a Multivariate Framework

More information

Game, set and match: The favorite-long shot bias in tennis betting exchanges.

Game, set and match: The favorite-long shot bias in tennis betting exchanges. Game, set and match: The favorite-long shot bias in tennis betting exchanges. Isabel Abinzano, Luis Muga and Rafael Santamaria Public University of Navarre (SPAIN) June 2015 Abstract: We test for the existence

More information

Factors Affecting the Probability of Arrests at an NFL Game

Factors Affecting the Probability of Arrests at an NFL Game Factors Affecting the Probability of Arrests at an NFL Game Patrick Brown 1. Introduction Every NFL season comes with its fair share of stories about rowdy fans taking things too far and getting themselves

More information

The Effects of Altitude on Soccer Match Outcomes

The Effects of Altitude on Soccer Match Outcomes The Effects of Altitude on Soccer Match Outcomes Tyler Williams Massachusetts Institute of Technology Cambridge, MA, United States, 02139 Email: tkwillia@mit.edu Abstract Soccer teams regularly compete

More information

Measuring Returns to Scale in Nineteenth-Century French Industry Technical Appendix

Measuring Returns to Scale in Nineteenth-Century French Industry Technical Appendix Measuring Returns to Scale in Nineteenth-Century French Industry Technical Appendix Ulrich Doraszelski Hoover Institution, Stanford University March 2004 Formal Derivations Gross-output vs value-added

More information

An Economic Analysis of Spectator Demand, Club Performance, and Revenue Sharing in English Premier League Football.

An Economic Analysis of Spectator Demand, Club Performance, and Revenue Sharing in English Premier League Football. An Economic Analysis of Spectator Demand, Club Performance, and Revenue Sharing in English Premier League Football. Adam John Cox The thesis is submitted in partial fulfilment of the requirements for the

More information

Behavior under Social Pressure: Empty Italian Stadiums and Referee Bias

Behavior under Social Pressure: Empty Italian Stadiums and Referee Bias Behavior under Social Pressure: Empty Italian Stadiums and Referee Bias Per Pettersson-Lidbom a and Mikael Priks bc* April 11, 2010 Abstract Due to tightened safety regulation, some Italian soccer teams

More information

Going with your Gut: The (In)accuracy of Forecast Revisions in a Football Score Prediction Game

Going with your Gut: The (In)accuracy of Forecast Revisions in a Football Score Prediction Game Going with your Gut: The (In)accuracy of Forecast Revisions in a Football Score Prediction Game Carl Singleton University of Reading J. James Reade University of Reading Alasdair Brown University of East

More information

Analysis of performance at the 2007 Cricket World Cup

Analysis of performance at the 2007 Cricket World Cup Analysis of performance at the 2007 Cricket World Cup Petersen, C., Pyne, D.B., Portus, M.R., Cordy, J. and Dawson, B Cricket Australia, Department of Physiology, Australian Institute of Sport, Human Movement,

More information

Economics of Sport (ECNM 10068)

Economics of Sport (ECNM 10068) Economics of Sport (ECNM 10068) Lecture 7: Theories of a League - Part I Carl Singleton 1 The University of Edinburgh 6th March 2018 1 Carl.Singleton@ed.ac.uk 1 / 27 2 / 27 Theories of a League - Part

More information

Determinants of college hockey attendance

Determinants of college hockey attendance University of Northern Iowa UNI ScholarWorks Honors Program Theses University Honors Program 2014 Determinants of college hockey attendance Max Martino University of Northern Iowa Copyright 2014 Max Martino

More information

Statistical Modeling of Consumers Participation in Gambling Markets and Frequency of Gambling

Statistical Modeling of Consumers Participation in Gambling Markets and Frequency of Gambling Statistical Modeling of Consumers Participation in Gambling Markets and Frequency of Gambling Brad R. Humphreys University of Alberta Department of Economics Yang Seung Lee University of Alberta Department

More information

High Bets for the Win?

High Bets for the Win? High Bets for the Win? The Role of Stake Size in Sports Betting Radek Janhuba & Jakub Mikulka (CERGE-EI) June 4, 2018 Latest version: http://www.radekjanhuba.com/files/rjjm_stakes.pdf Abstract We analyze

More information

Why We Should Use the Bullpen Differently

Why We Should Use the Bullpen Differently Why We Should Use the Bullpen Differently A look into how the bullpen can be better used to save runs in Major League Baseball. Andrew Soncrant Statistics 157 Final Report University of California, Berkeley

More information

Is the League of Ireland Under supported? Evidence from a European Perspective* Raymond Bachan and Barry Reilly

Is the League of Ireland Under supported? Evidence from a European Perspective* Raymond Bachan and Barry Reilly IstheLeagueofIrelandUndersupported? EvidencefromaEuropeanPerspective* RaymondBachanandBarryReilly Authors Affiliation: Bachan:BrightonBusinessSchool,UniversityofBrighton,Brighton,BN24AT Reilly: Department

More information

International Discrimination in NBA

International Discrimination in NBA International Discrimination in NBA Sports Economics Drew Zhong Date: May 7 2017 Advisory: Prof. Benjamin Anderson JEL Classification: L83; J31; J42 Abstract Entering the 21st century, the National Basketball

More information

THE YANKEES EFFECT: THE IMPACT OF INTERLEAGUE PLAY AND THE UNBALANCED SCHEDULE ON MAJOR LEAGUE BASEBALL ATTENDANCE

THE YANKEES EFFECT: THE IMPACT OF INTERLEAGUE PLAY AND THE UNBALANCED SCHEDULE ON MAJOR LEAGUE BASEBALL ATTENDANCE THE YANKEES EFFECT: THE IMPACT OF INTERLEAGUE PLAY AND THE UNBALANCED SCHEDULE ON MAJOR LEAGUE BASEBALL ATTENDANCE Rodney J. Paul*, Andrew P. Weinbach**, and Peter C. Melvin*** ABSTRACT Major League Baseball

More information

Volume 37, Issue 3. Elite marathon runners: do East Africans utilize different strategies than the rest of the world?

Volume 37, Issue 3. Elite marathon runners: do East Africans utilize different strategies than the rest of the world? Volume 37, Issue 3 Elite marathon runners: do s utilize different strategies than the rest of the world? Jamie Emerson Salisbury University Brian Hill Salisbury University Abstract This paper investigates

More information

On the advantage of serving first in a tennis set: four years at Wimbledon

On the advantage of serving first in a tennis set: four years at Wimbledon Published as: The Statistician (1999), 48, 247-256. On the advantage of serving first in a tennis set: four years at Wimbledon By Jan R. Magnus 1 CentER, Tilburg University, P.O. Box 90153, 5000 LE Tilburg,

More information

Swinburne Research Bank

Swinburne Research Bank Swinburne Research Bank http://researchbank.swinburne.edu.au Combining player statistics to predict outcomes of tennis matches. Tristan Barnet & Stephen R. Clarke. IMA journal of management mathematics

More information

An Analysis of Factors Contributing to Wins in the National Hockey League

An Analysis of Factors Contributing to Wins in the National Hockey League International Journal of Sports Science 2014, 4(3): 84-90 DOI: 10.5923/j.sports.20140403.02 An Analysis of Factors Contributing to Wins in the National Hockey League Joe Roith, Rhonda Magel * Department

More information

#1 Accurately Rate and Rank each FBS team, and

#1 Accurately Rate and Rank each FBS team, and The goal of the playoffpredictor website is to use statistical analysis in the absolute simplest terms to: #1 Accurately Rate and Rank each FBS team, and #2 Predict what the playoff committee will list

More information

Pierce 0. Measuring How NBA Players Were Paid in the Season Based on Previous Season Play

Pierce 0. Measuring How NBA Players Were Paid in the Season Based on Previous Season Play Pierce 0 Measuring How NBA Players Were Paid in the 2011-2012 Season Based on Previous Season Play Alex Pierce Economics 499: Senior Research Seminar Dr. John Deal May 14th, 2014 Pierce 1 Abstract This

More information

Chapter 12 Practice Test

Chapter 12 Practice Test Chapter 12 Practice Test 1. Which of the following is not one of the conditions that must be satisfied in order to perform inference about the slope of a least-squares regression line? (a) For each value

More information

Player dismissal and full time results in the UEFA

Player dismissal and full time results in the UEFA Player dismissal and full time results in the UEFA Champions League and Europa League Arne De Meyere, a Ward Vanruymbeke, a and Stijn Baert a,b,c,d,e,f,g * aghent University, Ghent, Belgium; b Research

More information

Real-Time Electricity Pricing

Real-Time Electricity Pricing Real-Time Electricity Pricing Xi Chen, Jonathan Hosking and Soumyadip Ghosh IBM Watson Research Center / Northwestern University Yorktown Heights, NY, USA X. Chen, J. Hosking & S. Ghosh (IBM) Real-Time

More information

Robust specification testing in regression: the FRESET test and autocorrelated disturbances

Robust specification testing in regression: the FRESET test and autocorrelated disturbances Robust specification testing in regression: the FRESET test and autocorrelated disturbances Linda F. DeBenedictis and David E. A. Giles * Policy and Research Division, Ministry of Human Resources, 614

More information

Analyses of the Scoring of Writing Essays For the Pennsylvania System of Student Assessment

Analyses of the Scoring of Writing Essays For the Pennsylvania System of Student Assessment Analyses of the Scoring of Writing Essays For the Pennsylvania System of Student Assessment Richard Hill The National Center for the Improvement of Educational Assessment, Inc. April 4, 2001 Revised--August

More information

Conference Call! NCAA Conference Realignment and Football Game Day Attendance. By: Mark Groza Spring 2007

Conference Call! NCAA Conference Realignment and Football Game Day Attendance. By: Mark Groza Spring 2007 Conference Call! NCAA Conference Realignment and Football Game Day Attendance. By: Mark Groza Spring 2007 Table of Contents Acknowledgments 3 Abstract 4 Introduction..5 Literature Review. 6 Empirical Model..

More information

Individual Behavior and Beliefs in Parimutuel Betting Markets

Individual Behavior and Beliefs in Parimutuel Betting Markets Mini-Conference on INFORMATION AND PREDICTION MARKETS Individual Behavior and Beliefs in Parimutuel Betting Markets Frédéric KOESSLER University of Cergy-Pontoise (France) Charles NOUSSAIR Faculty of Arts

More information

The Impact of Sports League Structure on Fan Welfare

The Impact of Sports League Structure on Fan Welfare The Impact of Sports League Structure on Fan Welfare Daniel Nussbaum Faculty Advisor: Professor Collin Raymond May 4, 2016 Submitted to the Department of Economics at Amherst College in partial fulfillment

More information

An economic analysis of attendance demand for One Day International cricket

An economic analysis of attendance demand for One Day International cricket An economic analysis of attendance demand for One Day International cricket Abhinav Sacheti, Ian Gregory-Smith and David Paton Abstract The future of One Day International (ODI) cricket has come under

More information

Working Paper No

Working Paper No Working Paper No. 2010-05 Prices, Point Spreads and Profits: Evidence from the National Football League Brad R. Humphreys University of Alberta February 2010 Copyright to papers in this working paper series

More information

Stats 2002: Probabilities for Wins and Losses of Online Gambling

Stats 2002: Probabilities for Wins and Losses of Online Gambling Abstract: Jennifer Mateja Andrea Scisinger Lindsay Lacher Stats 2002: Probabilities for Wins and Losses of Online Gambling The objective of this experiment is to determine whether online gambling is a

More information

Home Team Advantage in the NBA: The Effect of Fan Attendance on Performance

Home Team Advantage in the NBA: The Effect of Fan Attendance on Performance MPRA Munich Personal RePEc Archive Home Team Advantage in the NBA: The Effect of Fan on Performance Vincent La Dartmouth College 14 March 2014 Online at https://mpra.ub.uni-muenchen.de/54579/ MPRA Paper

More information

An exploration of the origins of the favourite-longshot bias in horserace betting markets

An exploration of the origins of the favourite-longshot bias in horserace betting markets An exploration of the origins of the favourite-longshot bias in horserace betting markets Alistair Bruce, Johnnie Johnson & Jiejun Yu Centre for Risk Research University of Southampton 1 University of

More information

Navigate to the golf data folder and make it your working directory. Load the data by typing

Navigate to the golf data folder and make it your working directory. Load the data by typing Golf Analysis 1.1 Introduction In a round, golfers have a number of choices to make. For a particular shot, is it better to use the longest club available to try to reach the green, or would it be better

More information

Journal of Sports Economics 2000; 1; 299

Journal of Sports Economics 2000; 1; 299 Journal of Sports Economics http://jse.sagepub.com Equal Pay on the Hardwood: The Earnings Gap Between Male and Female NCAA Division I Basketball Coaches Brad R. Humphreys Journal of Sports Economics 2000;

More information

Existence of Nash Equilibria

Existence of Nash Equilibria Existence of Nash Equilibria Before we can prove the existence, we need to remind you of the fixed point theorem: Kakutani s Fixed Point Theorem: Consider X R n a compact convex set and a function f: X

More information

The Impact of Star Power and Team Quality on NBA Attendance THESIS

The Impact of Star Power and Team Quality on NBA Attendance THESIS The Impact of Star Power and Team Quality on NBA Attendance THESIS Presented in Partial Fulfillment of the Requirements for the Honors Research Distinction in the Fisher College of Business at The Ohio

More information

When Falling Just Short is a Good Thing: the Effect of Past Performance on Improvement.

When Falling Just Short is a Good Thing: the Effect of Past Performance on Improvement. Title: When Falling Just Short is a Good Thing: the Effect of Past Performance on Improvement. Mariya Burdina (corresponding author) Associate Professor of Economics University of Central Oklahoma 100

More information

MODELLING THE EFFECT OF HEALTH RISK PERCEPTION ON

MODELLING THE EFFECT OF HEALTH RISK PERCEPTION ON MODELLING THE EFFECT OF HEALTH RISK PERCEPTION ON PREFERENCES AND CHOICE SET FORMATION OVER TIME: RECREATIONAL HUNTING SITE CHOICE AND CHRONIC WASTING DISEASE THUY TRUONG VIC ADAMOWICZ PETER BOXALL Resource

More information

Using Markov Chains to Analyze a Volleyball Rally

Using Markov Chains to Analyze a Volleyball Rally 1 Introduction Using Markov Chains to Analyze a Volleyball Rally Spencer Best Carthage College sbest@carthage.edu November 3, 212 Abstract We examine a volleyball rally between two volleyball teams. Using

More information

The relationship between payroll and performance disparity in major league baseball: an alternative measure. Abstract

The relationship between payroll and performance disparity in major league baseball: an alternative measure. Abstract The relationship between payroll and performance disparity in major league baseball: an alternative measure Daniel Mizak Frostburg State University Anthony Stair Frostburg State University Abstract This

More information

Journal of Human Sport and Exercise E-ISSN: Universidad de Alicante España

Journal of Human Sport and Exercise E-ISSN: Universidad de Alicante España Journal of Human Sport and Exercise E-ISSN: 1988-5202 jhse@ua.es Universidad de Alicante España SOÓS, ISTVÁN; FLORES MARTÍNEZ, JOSÉ CARLOS; SZABO, ATTILA Before the Rio Games: A retrospective evaluation

More information

Predicting the Total Number of Points Scored in NFL Games

Predicting the Total Number of Points Scored in NFL Games Predicting the Total Number of Points Scored in NFL Games Max Flores (mflores7@stanford.edu), Ajay Sohmshetty (ajay14@stanford.edu) CS 229 Fall 2014 1 Introduction Predicting the outcome of National Football

More information

Using Poisson Distribution to predict a Soccer Betting Winner

Using Poisson Distribution to predict a Soccer Betting Winner Using Poisson Distribution to predict a Soccer Betting Winner By SYED AHMER RIZVI 1511060 Section A Quantitative Methods - I APPLICATION OF DESCRIPTIVE STATISTICS AND PROBABILITY IN SOCCER Concept This

More information

Game Theory (MBA 217) Final Paper. Chow Heavy Industries Ty Chow Kenny Miller Simiso Nzima Scott Winder

Game Theory (MBA 217) Final Paper. Chow Heavy Industries Ty Chow Kenny Miller Simiso Nzima Scott Winder Game Theory (MBA 217) Final Paper Chow Heavy Industries Ty Chow Kenny Miller Simiso Nzima Scott Winder Introduction The end of a basketball game is when legends are made or hearts are broken. It is what

More information

In Search of Sure Things An Analysis of Market Efficiency in the NFL Betting Market

In Search of Sure Things An Analysis of Market Efficiency in the NFL Betting Market In Search of Sure Things An Analysis of Market Efficiency in the NFL Betting Market By: Rick Johnson Advisor: Professor Eric Schulz Submitted for Honors in Mathematical Methods in the Social Sciences Spring,

More information

ECO 199 GAMES OF STRATEGY Spring Term 2004 Precept Materials for Week 3 February 16, 17

ECO 199 GAMES OF STRATEGY Spring Term 2004 Precept Materials for Week 3 February 16, 17 ECO 199 GAMES OF STRATEGY Spring Term 2004 Precept Materials for Week 3 February 16, 17 Illustration of Rollback in a Decision Problem, and Dynamic Games of Competition Here we discuss an example whose

More information

Distancei = BrandAi + 2 BrandBi + 3 BrandCi + i

Distancei = BrandAi + 2 BrandBi + 3 BrandCi + i . Suppose that the United States Golf Associate (USGA) wants to compare the mean distances traveled by four brands of golf balls when struck by a driver. A completely randomized design is employed with

More information

An Application of Signal Detection Theory for Understanding Driver Behavior at Highway-Rail Grade Crossings

An Application of Signal Detection Theory for Understanding Driver Behavior at Highway-Rail Grade Crossings An Application of Signal Detection Theory for Understanding Driver Behavior at Highway-Rail Grade Crossings Michelle Yeh and Jordan Multer United States Department of Transportation Volpe National Transportation

More information

Golfers in Colorado: The Role of Golf in Recreational and Tourism Lifestyles and Expenditures

Golfers in Colorado: The Role of Golf in Recreational and Tourism Lifestyles and Expenditures Golfers in Colorado: The Role of Golf in Recreational and Tourism Lifestyles and Expenditures by Josh Wilson, Phil Watson, Dawn Thilmany and Steve Davies Graduate Research Assistants, Associate Professor

More information

The impacts of football point systems on the competitive balance: evidence from some European footbal leagues

The impacts of football point systems on the competitive balance: evidence from some European footbal leagues MPRA Munich Personal RePEc Archive The impacts of football point systems on the competitive balance: evidence from some European footbal leagues Ferda HALICIOGLU Department of Economics, Yeditepe University

More information

Equation 1: F spring = kx. Where F is the force of the spring, k is the spring constant and x is the displacement of the spring. Equation 2: F = mg

Equation 1: F spring = kx. Where F is the force of the spring, k is the spring constant and x is the displacement of the spring. Equation 2: F = mg 1 Introduction Relationship between Spring Constant and Length of Bungee Cord In this experiment, we aimed to model the behavior of the bungee cord that will be used in the Bungee Challenge. Specifically,

More information

Are Workers Rewarded for Inconsistent Performance?

Are Workers Rewarded for Inconsistent Performance? Are Workers Rewarded for Inconsistent Performance? Anil Özdemir (University of Zurich) Giambattista Rossi (Birkbeck College) Rob Simmons (Lancaster University) Following Lazear (1995) a small body of personnel

More information

Basic Wage for Soccer Players in Japan :

Basic Wage for Soccer Players in Japan : The Otemon Journal of Australian Studies, vol., pp., Basic Wage for Soccer Players in Japan : Individual Performance and Profit Sharing Components Shimono Keiko Nagoya City University Abstract This paper

More information

An Analysis of the Effects of Long-Term Contracts on Performance in Major League Baseball

An Analysis of the Effects of Long-Term Contracts on Performance in Major League Baseball An Analysis of the Effects of Long-Term Contracts on Performance in Major League Baseball Zachary Taylor 1 Haverford College Department of Economics Advisor: Dave Owens Spring 2016 Abstract: This study

More information

Does the Three-Point Rule Make Soccer More Exciting? Evidence from a Regression Discontinuity Design

Does the Three-Point Rule Make Soccer More Exciting? Evidence from a Regression Discontinuity Design MPRA Munich Personal RePEc Archive Does the Three-Point Rule Make Soccer More Exciting? Evidence from a Regression Discontinuity Design Yoong Hon Lee and Rasyad Parinduri Nottingham University Business

More information

Monopsony Exploitation in Professional Sport: Evidence from Major League Baseball Position Players,

Monopsony Exploitation in Professional Sport: Evidence from Major League Baseball Position Players, Department of Economics Working Paper Series Monopsony Exploitation in Professional Sport: Evidence from Major League Baseball Position Players, 2000-2011 Brad R. Humphreys Hyunwoong Pyun Working Paper

More information

Running head: DATA ANALYSIS AND INTERPRETATION 1

Running head: DATA ANALYSIS AND INTERPRETATION 1 Running head: DATA ANALYSIS AND INTERPRETATION 1 Data Analysis and Interpretation Final Project Vernon Tilly Jr. University of Central Oklahoma DATA ANALYSIS AND INTERPRETATION 2 Owners of the various

More information

Emergence of a professional sports league and human capital formation for sports: The Japanese Professional Football League.

Emergence of a professional sports league and human capital formation for sports: The Japanese Professional Football League. MPRA Munich Personal RePEc Archive Emergence of a professional sports league and human capital formation for sports: The Japanese Professional Football League. Eiji Yamamura 26. February 2013 Online at

More information

Section I: Multiple Choice Select the best answer for each problem.

Section I: Multiple Choice Select the best answer for each problem. Inference for Linear Regression Review Section I: Multiple Choice Select the best answer for each problem. 1. Which of the following is NOT one of the conditions that must be satisfied in order to perform

More information

DOES THE NUMBER OF DAYS BETWEEN PROFESSIONAL SPORTS GAMES REALLY MATTER?

DOES THE NUMBER OF DAYS BETWEEN PROFESSIONAL SPORTS GAMES REALLY MATTER? Trevor Hardy Canadian Football League Toronto, ON Canada Keith A. Willoughby Edwards School of Business University of Saskatchewan DOES THE NUMBER OF DAYS BETWEEN PROFESSIONAL SPORTS GAMES REALLY MATTER?

More information

How predictable are the FIFA worldcup football outcomes? An empirical analysis

How predictable are the FIFA worldcup football outcomes? An empirical analysis Applied Economics Letters, 2008, 15, 1171 1176 How predictable are the FIFA worldcup football outcomes? An empirical analysis Saumik Paul* and Ronita Mitra Department of Economics, Claremont Graduate University,

More information

Professionals Play Minimax: Appendix

Professionals Play Minimax: Appendix Professionals Play Minimax: Appendix Ignacio Palacios-Huerta Brown University July 2002 In this supplementary appendix we first report additional descriptive statistics in section 1. We begin by reporting

More information

Effects of Incentives: Evidence from Major League Baseball. Guy Stevens April 27, 2013

Effects of Incentives: Evidence from Major League Baseball. Guy Stevens April 27, 2013 Effects of Incentives: Evidence from Major League Baseball Guy Stevens April 27, 2013 1 Contents 1 Introduction 2 2 Data 3 3 Models and Results 4 3.1 Total Offense................................... 4

More information

NBA TEAM SYNERGY RESEARCH REPORT 1

NBA TEAM SYNERGY RESEARCH REPORT 1 NBA TEAM SYNERGY RESEARCH REPORT 1 NBA Team Synergy and Style of Play Analysis Karrie Lopshire, Michael Avendano, Amy Lee Wang University of California Los Angeles June 3, 2016 NBA TEAM SYNERGY RESEARCH

More information

Regression to the Mean at The Masters Golf Tournament A comparative analysis of regression to the mean on the PGA tour and at the Masters Tournament

Regression to the Mean at The Masters Golf Tournament A comparative analysis of regression to the mean on the PGA tour and at the Masters Tournament Regression to the Mean at The Masters Golf Tournament A comparative analysis of regression to the mean on the PGA tour and at the Masters Tournament Kevin Masini Pomona College Economics 190 2 1. Introduction

More information